Madison's Morning Memo

Archives
Log in
Subscribe
October 30, 2026

Notion Rebuilds AI Pricing Around Credits, Not Seats

AI PM Research

Friday, October 30, 2026

Notion Rebuilds AI Pricing Around Credits, Not Seats

Notion rebuilt its AI pricing around a credit system while LangChain kept shipping pieces of what it calls "loop engineering," and together they're both answering the same question: once agents start running for minutes or hours instead of seconds, how do you price and monitor something whose cost structure you can't fully predict upfront?


Notion - Rebuilds AI Model Routing and Moves to a Credit-Based Pricing System

Notion shipped a reworked AI system described as "agents, rebuilt to do more for less" — co-authored by co-founder Akshay Kotha. The update bundles smarter model routing, long-running agents, bring-your-own-AI (BYOAI) options, and a new credit-based pricing system, replacing whatever flat or seat-based AI allowance existed before.

LangChain - "The Art of Loop Engineering" Plus Browser Use for Long-Running Agents

LangChain published a framework it calls "loop engineering" — the idea that reliable agents require more than a good model; they need a carefully designed harness around the core agent loop, built for the specific task, with instrumentation at every level. 10 added browser use (agents can now operate on sites with no API and hand control back to a human for sign-in) and async subagents for background work.

Instinct - Group Chat Agents on iMessage and User-Set Spending Budgets

Noah Shinn, founder of Instinct, announced two product changes in quick succession. First: Instinct in group chats is now available to everyone on iMessage, with a specific trust design — the group gets its own shared Instinct instance that only knows what's in that chat, not any individual's personal accounts or data, and if it needs something from an individual it routes through that person's personal Instinct for confirmation first.

Also today

  • Simon Willison: Flags a quote from Ben Hylak on agent/model behavior worth reading alongside his GitHub Octoverse 2026 writeup and notes on ChatGPT's new API credits benefit for eligible subscriptions
  • OpenAI: Published an updated Model Spec defining five core values ("sound judgment, human agency, and faithfulness to human intent") for more capable AI, alongside a post on how it detects frontier model misalignment via layered monitoring
  • Anthropic: Published "Early progress on accelerating drug development with Claude," describing deep research partnerships with six organizations developing new medicines over the past several months
  • ElevenLabs: Shipped a "fast LLM service tier" for ElevenAgents, trading a per-token price premium for higher throughput and lower latency, configurable per-agent
  • Figma: Shipped Google Drive connectivity for the Figma agent and Figma Make on all plans, plus internal publishing for Weave tools on Organization/Enterprise plans
Read the full update →

AI PM Research · Daily updates on AI products and strategy

Don't miss what's next. Subscribe to Madison's Morning Memo:
← Newer Notion's Agent Overhaul and the Cost Transparency Shift Older → GitHub Universe, Notion's Credit Reset, and Agent Harness Design
ai-pm-research.netlify.app
Powered by Buttondown, the easiest way to start and grow your newsletter.